Materials

Horizons
View Article Online
COMMUNICATION View Journal | View Issue

Integrated bioelectronic proton-gated logic

elements utilizing nanoscale patterned Nafion†
Cite this: Mater. Horiz., 2021,
8, 224
J. G. Gluschke, a J. Seidl,a R. W. Lyttleton,ab K. Nguyen,a M. Lagier,a F. Meyer,a
Received 1st July 2020, P. Krogstrup,c J. Nygård,c S. Lehmann,b A. B. Mostert,d P. Meredithef and
Accepted 30th October 2020
A. P. Micolich *a
Published on 02 November 2020. Downloaded on 9/15/2023 5:20:46 AM.

DOI: 10.1039/d0mh01070g

rsc.li/materials-horizons

A central endeavour in bioelectronics is the development of logic

elements to transduce and process ionic to electronic signals. New concepts
Motivated by this challenge, we report fully monolithic, nanoscale Nafion is an ionomer that is heavily used as a proton conductor in proton
exchange membrane fuel cells due to its exceptionally high protonic
logic elements featuring n- and p-type nanowires as electronic
conductivity. Nafion has recently generated considerable excitement as a
channels that are proton-gated by electron-beam patterned potential component in artificial synapse devices. A major roadblock to
Nafion. We demonstrate inverter circuits with state-of-the-art these new applications is the lack of methods for scalable patterning of
ion-to-electron transduction performance giving DC gain exceed- Nafion films into device structures and integrated circuits. Our new
ing 5 and frequency response up to 2 kHz. A key innovation concept here is the demonstration of a scalable method for electron-
beam processing of Nafion films spin-coated on solid substrates to give
facilitating the logic integration is a new electron-beam process
nanoscale patterns with linewidths as low as 125 nm. This new Nafion
for patterning Nafion with linewidths down to 125 nm. This process patterning capability enables its use as an ion-conducting element in
delivers feature sizes compatible with low voltage, fast switching complex nanoscale devices and integrated circuits in bioelectronics. We
elements. This expands the scope for Nafion as a versatile pattern- provide a proof-of-concept demonstration by producing a hybrid organic–
able high-proton-conductivity element for bioelectronics and inorganic ion-to-electron transducer circuit giving DC gain exceeding 5
and frequency response up to 2 kHz. This performance competes well
other applications requiring nanoengineered protonic membranes
with state-of-the-art ion-to-electron transducers featuring PEDOT:PSS.
and electrodes. Our work highlights the outstanding potential of Nafion as a high-
speed device-level ion-transport material for advanced bioelectronics
applications.

Introduction
bioelectronics is the diversity of charge carriers – in addition to
A key challenge in the emerging discipline of bioelectronics is the familiar electrons and holes that carry charge in solid-state
the connection and integration of read–write-process electronic electronic materials, a large cast of ions, e.g., H+, K+, Na+,
circuitry directly or indirectly with a target biological entity.1–3 Cl , etc., are the signal carrying entities in biology.1,5 Thus, a
The field sits at the nexus between the physical, chemical and key challenge for the field is the efficient and effective inter-
life sciences, and holds considerable promise for the creation conversion between electronic and ionic signals, a process
of disruptive technologies for healthcare.4 Indeed, some claim commonly referred to as ‘transduction’. This challenge,
that bioelectronics, and related endeavours such as bionics, are although simply stated, is far from trivial since the underlying
the next evolutionary phases of medicine.4 A defining feature of physics of electrons, being quantum mechanical in nature, is
very different to the large, center of mass, classical behaviour
a
School of Physics, University of New South Wales, Sydney, NSW 2052, Australia.
exhibited by ions.
E-mail: adam.micolich@nanoelectronics.physics.unsw.edu.au
b
NanoLund, Lund University, SE-221 00 Lund, Sweden
The requirements for practical bioelectronic applications
c
Center for Quantum Devices and Station Q Copenhagen, University of are wide-ranging and include: biocompatibility, high electronic
Copenhagen, DK-2100 Copenhagen, Denmark and ionic conductivity, long lifetime and stability under biolo-
d
Chemistry Department, Swansea University, Swansea SA2 8PP, Wales, UK gical conditions, as well as respectable prospects for patterning,
e
Physics Department, Swansea University, Swansea SA2 8PP, Wales, UK
f
device fabrication and deployment on soft conformable
School of Mathematics and Physics, University of Queensland, Brisbane,
QLD 4072, Australia
substrates.2,6,7 These requirements have driven the exploration
† Electronic supplementary information (ESI) available. See DOI: 10.1039/ and development of a vast array of soft, organic, ionically active
d0mh01070g materials. These are then incorporated into device architectures

224 | Mater. Horiz., 2021, 8, 224--233 This journal is © The Royal Society of Chemistry 2021
 View Article Online

Communication Materials Horizons

by considering complexities such as capacitive versus faradaic it highlights a major challenge for Nafion as a device material –
effects,8 and device operating concepts, a classic example being scalability and integration. We developed a process whereby
the choice between organic electrochemical transistors (OECTs) Nafion can be used as a negative electron-beam lithography
and organic field-effect transistors (OFETs).9 Of the various resist to enable patterned structures with lateral dimensions as
bioelectronic materials, PEDOT–PSS is the most developed owing small as 125 nm to address this challenge. We show that Nafion
to its long history as an electronic material10 and recent develop- has sufficient solvent-resistance to tolerate subsequent litho-
ment towards neural interfacing applications.1,11 However, neural graphic steps, giving a major versatility advantage compared to
interfacing is only one part of a greater bioelectronic mission that many other ion-transport materials used in bioelectronics.
includes the development of protonic transistors,12,13 ionic logic14 For our electronic componentry, we use a III–V nanowire
and ‘iontronic’ devices,15 biosymbiotic circuitry,16 e.g., ‘electronic complementary circuit platform24 developed such that the
plants’,17 synapse-like devices,18,19 pH sensor and control entire Nafion-plus-III–V architecture is monolithically inte-
systems,20–22 and biologically-tunable light-gated transistors.23 grated onto a single chip. We demonstrate an inverter circuit
Advances in electronics are not only driven by new materials with DC gain exceeding 5 and AC frequency response up to
and new developments in existing materials, but also by new 2 kHz for sine wave signals and hundreds of Hz for square wave
Published on 02 November 2020. Downloaded on 9/15/2023 5:20:46 AM.

materials combinations. An example is a recent shift from signals. This frequency response is sufficient for neural sensing
all-organic or all-inorganic systems towards more hybrid applications, where the oscillation spectrum for brain activity
organic–inorganic systems, where: (a) identified deficiencies ranges from 1–200 Hz.11 Response beyond 200 Hz is desirable,
of organics, e.g., speed/solubility, are resolved using traditional however, as it affords better signal fidelity. This has driven a
inorganic semiconductors such as Si or III–Vs3,21,23,24 or (b) race to produce high-sensitivity ion-to-electron transducers that
deficiencies of inorganics, e.g., slow ion diffusion, are resolved operate up into the kHz, a feat that has only very recently been
using organic ionomers such as Nafion.19 For example, OECTs achieved with PEDOT:PSS OECTs.31
have high transconductances25,26 but can lose their perfor- An important aspect to note on the novelty of this work is
mance advantage to OFETs at higher frequencies.9 The carrier that we use Nafion itself as the electron-beam lithography
mobility of organic semiconductors is also several orders of resist to leave residual nanoscale Nafion structures that are
magnitude lower than for their inorganic counterparts, further subsequently used as ionically active elements for device
limiting electronic performance, e.g., switching speed. Organic applications. The Nafion acts as a negative tone resist, opposite
semiconductors are predominantly p-type and the relative to the more conventionally used positive resist polymethyl-
lack of high performance n-type materials makes all-organic methacrylate (PMMA) used in electron-beam lithography.
complementary circuits suitable for bioelectronics applications The negative tone means we only need to scan the regions
rare.27 In this regard we note that complementary logic we wish to retain, such that scalability depends only on the
would be the most likely platform for integrated bioelectronic total area of the Nafion features we wish to keep not the total
circuitry, and so the scarcity of n-type organics is problematic substrate area. Our work is also very different, in both method
and an area of considerable activity in the synthetic chemistry and intended application, to previous work where freestanding
community.28 Sub-micron patterning can also be difficult for commercial Nafion-117 membranes were patterned with nano-
many organics due to incompatibilities with the solvents scale holes using either conventional PMMA-based electron-
used in lithographic processing, thereby limiting ultimate beam lithography32,33 or high-energy electron-beam damage34
device miniaturization and integration. That said, traditional to influence ion permeability for proton exchange membrane
inorganic semiconductors on their own are also unsuited to (PEM) fuel cell applications.
bioelectronics, showing poor ion permeability, problems with Our work here represents a major step forward on prior-art
surface effects and dangling bonds, and insufficient mechanical for both Nafion in a device context18,19 and hybrid organic–
flexibility – let alone potential cytotoxicity if intimately exposed inorganic nanowire transducers.24 It represents what we believe
at an interface.7 to be the first integrated logic circuit capable of processing ion
Here we report a hybrid organic–inorganic system that and electron signals simultaneously and an interesting alter-
draws a series of ‘best in category’ materials together into a native to PEDOT:PSS OECTs for bioelectronics operational into
single, monolithically integrated, nanoscale-patterned ion-to- the kHz frequency regime. Our results also highlight the
electron transduction platform with strong prospects for potential for Nafion as a nanoscale patterned material for use
bioelectronic applications. For our ion-transport material we as an ionically functional element in other bioelectronic device
chose the synthetic polymer Nafion, which has an exceptionally architectures, or indeed as a structured proton membrane or
high protonic conductivity (78 mS cm 1),29 more than an order electrode for energy storage applications.
of magnitude higher than other biopolymers.30 Nafion has
previously been used in artificial synapse devices.18,19 These
macroscopic devices were made by sandwiching either a Results and discussion
0.18 mm thick Nafion-117 membrane19 or cellulose tissue
wipe soaked in Nafion solution18 between the two halves of Direct electron-beam patterning of Nafion
the device or spin-coating a Nafion film over a prepatterned Fig. 1A–C illustrate the approach used for direct patterning of a
epoxy (SU-8) mask.18 Although effective for proof-of-concept, Nafion thin-film by electron-beam exposure. The thin-film of

This journal is © The Royal Society of Chemistry 2021 Mater. Horiz., 2021, 8, 224--233 | 225
 View Article Online

Materials Horizons Communication

Nafion was produced by spin-coating commercially available far more versatile material from a fabrication perspective, as
5% Nafion-117 solution on a SiO2-on-n+–Si substrate (Fig. 1A). we will show below. The negative contrast of Nafion under
Select areas were exposed with a 5–10 keV electron beam to a electron-beam exposure is also fortuitous because it means that
dose of B10 mC cm 2 using a Raith150-Two electron-beam patterned regions can be written directly, making for a favour-
lithography system (Fig. 1B). This causes the electron-beam ably scalable process compared to, for example, a positive
exposed Nafion regions to become comparatively insoluble, contrast resist such as the polymethylmethacrylate (PMMA)
such that after brief immersion in a developer solution typically used in electron-beam lithography.
(1 : 1 2-propanol : acetone), the unexposed Nafion is dissolved On the issue of scaling, EBL systems are now routinely
away and only the exposed Nafion remains (Fig. 1C). In other used in large-scale fabrication of, e.g., high-end photonic
words, we show that Nafion acts directly as a negative-tone nanostructures and nanoimprint lithography masters,35 provid-
electron-beam lithography resist. ing substantial scaling potential for our Nafion patterning
Fig. 1D–F demonstrate patterning of a 230 nm thick Nafion approach. Additionally, the direct negative-tone patterning of
film produced by spin-coating undiluted 5% Nafion-117 Nafion requires significantly fewer processing steps than
solution at 3000 rpm for 30 s, followed by electron-beam exposure indirect patterning via etch masks, e.g., as used for controlled
Published on 02 November 2020. Downloaded on 9/15/2023 5:20:46 AM.

and development for 60 s in 1 : 1 2-propanol : acetone. The roughening of commercial Nafion membrane surfaces.32,33
electron-beam-exposed region is evident by optical microscopy This indirect patterning process is based on a positive tone
prior to development (Fig. 1D), due to a slight change in refractive EBL resist used to create a metal etch mask for oxygen plasma
index presumably arising from electron-beam induced material etching. This approach is not suitable for our devices because
changes, and this region clearly remains undissolved after devel- it would require EBL exposure of the entire wafer except for
opment (Fig. 1E). Notably, the electron-beam exposed regions are the nanoscale Nafion channels. Another surface roughening
resistant to further solvent exposure thereafter. Fig. 1F shows the approach, developed for Nafion membrane and based on
same film after immersions in hot acetone (60 1C) for 60 min, local destruction of material from high-energy electron beam
room-temperature 2-propanol for 60 min and room-temperature exposure,34 shares this limitation. While these alternative
deionized water for 60 min, simulating the conditions typically approaches do not translate well to the device level, they
found in the lift-off process for subsequent patterning of metal underline a broader interest in nanoscale Nafion patterning.
features. The Nafion features survive this treatment remarkably Returning to the device fabrication, thinner Nafion films
well, in stark contrast to other patternable ion-conducting can be obtained by thinning the 5% Nafion-117 solution in
polymers, e.g., polyethylene oxide.24 This is a significant out- ethanol,36 and Fig. 1I and J demonstrate that the same electron-
come because it means the Nafion deposition/patterning does beam patterning process works well also for 25 nm Nafion
not need to be the final process step, thereby making Nafion a films, giving improved linewidths. Fig. 1G and I show Atomic

Fig. 1 Nafion as a novel negative-tone electron-beam lithography resist (A–C) illustration of direct electron beam patterning process. (A) Nafion is
spin coated on the substrate. (B) Nafion channels are exposed with the electron beam (shaded area). (C) The unexposed Nafion is dissolved in the 1 : 1 2-
propanol : acetone developer leaving the exposed regions. (D–F) Optical microscopy images of a 230 nm thick Nafion film (D) immediately after
electron-beam exposure, (E) after development, and (F) after prolonged exposure (see text) to the solvents 2-propanol, acetone, and H2O demonstrating
the robustness of the electron-beam patterned Nafion. The substrate appears blue and the patterned Nafion purple. (G–J) Atomic Force Microscopy
(AFM) scans and height profiles for (G and H) 230 nm thick and (I and J) 25 nm thick Nafion films. The AFM scans in (G) and (I) were obtained from lines
with beam-patterned width 300 nm and pitch 1 mm for the 230 nm thick Nafion film and 1.5 mm for the 25 nm thick Nafion film. The height profiles
(H) and (J) were obtained from a separate exposure where we wrote single lines with beam-patterned width 100 nm to determine ultimate resolution.
The post-development full width at half maximum (FWHM) linewidths were: (G) 650 nm, (H) 420 nm, (I) 360 nm and (J) 125 nm.

226 | Mater. Horiz., 2021, 8, 224--233 This journal is © The Royal Society of Chemistry 2021
 View Article Online

Communication Materials Horizons

Force Microscopy (AFM) scans obtained for line arrays with degrees,40 making our thermal test difficult to reconcile to our
lithographic width 300 nm and 1 mm/1.5 mm pitch respectively processing. For example, we note that our electron-beam energy
for the 230/25 nm thick films. We obtain a full linewidth at and dose, at 10 keV and 10 mC cm 2, are low by normal
half maximum of 650 nm for 230 nm-thick Nafion and 360 nm electron-beam lithography standards and our individual point
for 25 nm-thick Nafion for these arrays. Fig. 1H and J show exposure times are short (B3 ms). At this stage, while electron-
height profiles obtained for single lines with lithographic width beam induced chemical changes to the Nafion are the logical
100 nm written to test ultimate resolution, giving a full line- origin of our solubility contrast in light of conventional processes
width at half maximum of 420 nm for 230 nm-thick Nafion and for electron-beam lithography and our XPS data, structural
125 nm for 25 nm-thick Nafion. Line broadening occurs in changes arising from local heating may also play some role.
most EBL resists due to electron back-scattering from the More detailed studies of how electron-beam exposure affect
substrate.37 The ‘teardrop’-shaped beam interaction volume37 the properties of our Nafion films will be reported separately.
will naturally give broader lines for the 230 nm films (Fig. 1G).
For adjacent lines to be ionically independent (i.e., the Nafion Managing Nafion acidity and the fabrication of monolithic
thickness reaches zero for a finite distance between lines) the nanowire proton-to-electron transducers
Published on 02 November 2020. Downloaded on 9/15/2023 5:20:46 AM.

separation presently needs to be approximately 300 nm and The sulfonic acid groups in Nafion form hydrophilic pores that
1200 nm respectively for adjacent lines made in 25 nm and strongly adsorb water, to which the sulfonate acid groups
230 nm thick Nafion films. We expect however that Nafion donate H+ ions, thereby providing Nafion with its exceptionally
patterning resolution could be further improved by carefully high protonic conductivity.38 This presents an interesting twist
optimizing exposure parameters such as beam energy, aper- from a device fabrication perspective, because it also means
ture, and dose, if required. Additionally, anisotropic etch that Nafion is a solid superacid catalyst,41 one that we found
methods, e.g., reactive ion etching, could be used to etch away (unexpectedly) to rapidly dissolve InAs nanowires if the nano-
the tails of the Gaussian line profile to ionically isolate adjacent wire is connected to a large reservoir of electrons, i.e., any
lines at the base if finer line-separations are required. amount of metal (see ESI,† Fig. S8–S11). This process is
The devices presented in this work only use individual Nafion essentially the equivalent of a metal-assisted chemical etch
channels rather than arrays. Ionic crosstalk due to channel (MAC-etch),42 but here implemented via a localized solid
overlap is therefore not an issue presently. organic acid. This may have interesting prospects separate to
The complete mechanism for the electron-beam produced this work, but for this project it was a challenge that we could
solubility contrast is unclear at this stage but could be a only overcome because of the good solvent-resistance of the
combination of two contributions: (a) direct chemical changes electron-beam processed Nafion described earlier. The Nafion
induced by the electron-beam, and (b) structural changes due needs to be patterned prior to any metal deposition if a
to local heating produced by electron-beam interactions. We nanowire is going to survive the fabrication process, which in
explore these effects using X-ray photoelectron spectroscopy turn means Nafion needs to be able to survive lithography
(XPS) and thermal annealing studies (see ESI†). processes. As we demonstrate below, the electron-beam
For the XPS study, we compared electron-beam processed processed Nafion can survive not just one, but three subsequent
and unprocessed Nafion films. The data indicates cleavage of metallization steps in completing the device.
ether moieties on the sulfonated pendant groups. Nafion is a Fig. 2A–D show the fabrication process we ultimately used to
polymer featuring a polytetrafluoroethylene backbone with produce fully monolithic proton-to-electron transducers using
sulfonated perfluoropolyether pendant groups.38 The insoluble a two-nanowire complementary circuit (electronic side) gated
fluorinated backbone is rendered soluble by the sulfonated by a nanoscale-patterned Nafion gate element (protonic side).
pendant groups, meaning that pendant cleavage should pro- We begin with a SiO2-insulated n+–Si substrate pre-patterned
duce a solubility contrast. Any associated crosslinking between with large-scale interconnects and nanoscale alignment markers
Nafion moieties arising thereafter would add further to the for accurate positioning. We use an micromanipulator43 to
contrast. We also note, the XPS data indicates that there is no place a small number of n-type InAs and p-type GaAs nanowires
change in the proportion of sulfonated groups to the rest of in each 100  100 mm2 device field (Fig. 2A). We then register
the polymer. This indicates that the proton concentration in the nanowire position and pattern aligned Nafion strips using
the processed Nafion should remain unchanged after electron the procedure described earlier, prior to any metal contacts
beam exposure. being added to the nanowires to prevent them being etched by
Nafion is known to undergo structural changes that affect the Nafion (Fig. 2B). Three rounds of electron-beam lithography
solubility39 and protonic conductivity29 by thermal annealing and vacuum metal evaporation were used to: (i) define Ni/Au
at temperatures around 100 1C. Our thermal annealing tests ohmic contacts to the InAs nanowire and interconnects
on the solubility of unexposed spin-coated Nafion films in (Fig. 2C); (ii) define AuBe ohmic contacts to the GaAs nanowire;
our developer solution and found that solubility is lost after and (iii) define Ti/Pd contacts to the Nafion gate strips (Fig. 2D).
a 10 minute bake at 100 1C, but is retained even after several The complete fabrication protocol is given in the Methods
hours of baking at 60 1C. It is difficult to measure the local section. As an option, we can make the Nafion gate-strip
heating generated during the electron-beam lithography contact out of Ni/Au and write it as part of the first lithography
process, but prior work suggests it is typically of order tens of step (Fig. 2C). We have tried both approaches and find little

This journal is © The Royal Society of Chemistry 2021 Mater. Horiz., 2021, 8, 224--233 | 227
 View Article Online

Materials Horizons Communication

Published on 02 November 2020. Downloaded on 9/15/2023 5:20:46 AM.

Fig. 2 Process for making monolithic proton-to-electron transducers using Nafion and nanowires (A) InAs and GaAs nanowires were transferred to a
pre-patterned SiO2-on-n+–Si device substrate, (B) Nafion is spin-coated and electron-beam patterned, (C and D) three thermal evaporation steps follow
to define (C) InAs ohmic contacts (Ni/Au), (D) GaAs ohmic contacts (AuBe) and interconnects and gate contact (Ti/Pd). Note that the Ti/Pd element can
be written as Ni/Au as part of (C), as discussed in the text. (E) Optical micrograph of a completed inverter circuit with relevant features indicated.

functional difference thus far. An optical micrograph of a nanowire transistor, since this a novel aspect of our work in
completed functional device is shown in Fig. 2E. The active itself. In Fig. 3A and B we plot the source–drain current ISD
region has a typical footprint of 30  60 mm2 (not including versus Nafion electrode voltage VG at different source–drain
interconnects to external instrumentation). This could voltages VSD at maximal hydration (Fig. 3A) and different
be further reduced by placing nanowires closer together using hydration levels for fixed source–drain voltage VSD (Fig. 3B) to
more accurate nanowire alignment procedures such as highlight an important functional aspect of these devices.
resist trench alignment.44,45 Note that completed devices Proton conductivity in Nafion is facilitated by absorbed water,
remain operational even after several months of storage in a and is thus strongly humidity dependent.29 All measurements
N2-glovebox and careful handling. were thus performed in a sealed chamber with gas inlet/outlet
lines that enable the atmosphere to be vacuum purged,
DC electrical characterization of Nafion-gated nanowire backfilled with nitrogen, and then hydrated by maintaining a
transistors and inverter circuits continuous flow of nitrogen via a water bubbler. Under steady-
Before examining the inverter circuit performance, we briefly state conditions, this gives a relative humidity exceeding 90%.
demonstrate functionality of an individual Nafion-gated InAs As Fig. 3B demonstrates, gate action drops off sharply away

Fig. 3 DC characterization of Nafion-gated InAs nanowire transistors and inverter circuits (A and B) Characterization of the current ISD vs. voltage VG
applied to the Nafion gate electrode at (A) five different source–drain biases VSD = 10, 20, 30, 40, 50 mV under maximum hydration conditions (relative
humidity 495%) and (B) at VSD = 50 mV for relative humidities RH = 50% (yellow), 70% (orange) and 90% (blue). (C) Inverter circuit diagram and (D)
schematic of operation. An input signal Vin modulates the electrical double-layers in the Nafion, which affects the nanowire conductivity producing an
inverted copy of the input signal as the output Vout. (E) Output voltage Vout vs. input voltage Vin and (F) DC gain qVout/qVin vs. Vin under quasi-static (DC)
operation for the inverter circuit at maximum hydration (RH 4 95%). Data in (E and F) were obtained for four different drive voltages VDD = 250 mV
(yellow), 500 mV (orange), 750 mV (blue) and 1 V (green).

228 | Mater. Horiz., 2021, 8, 224--233 This journal is © The Royal Society of Chemistry 2021
 View Article Online

Communication Materials Horizons

from maximal hydration, indeed under ambient conditions this drives Vout towards ground (VDD), such that an inverted AC
our devices perform poorly if at all. This should not be viewed output signal results when an AC input is supplied.
as a major problem – bioelectronics implicitly involves wet Turning to the complete inverter device, in Fig. 3E we plot
environments, if anything, our devices at maximum hydration the output voltage Vout versus the input voltage Vin for four
are operated in a drier state than would be found in any different drive voltages VDD. Sweeping Vin between 0.3 V and
bioelectronic application. The data in Fig. 3A shows that these +0.3 V is enough to modulate Vout over 94% of VDD, i.e., between
devices perform well under maximal hydration conditions 2% and 96% of VDD. This indicates that the parasitic resistance
with sub-threshold slopes of typically 100–300 mV dec 1 in our inverter circuit is low, i.e., the dominant resistance
and on–off ratios of 103 and close to 104 in our best devices. contributions are from gateable channel material. In Fig. 3F
The operating range and performance is competitive with we plot the DC gain qVout/qVin versus Vin, with a 9-point smooth
existing liquid electrolyte-gated nanowire transistors,46–48 and applied post differentiation to prevent point-to-point derivative
excellent against equivalent all organic solid-state transducing noise from obscuring the signal. We obtain DC gain as high as
transistors.5,49 A more extensive characterization of a Nafion- 6, which compares well with the maximum gain recently
gated nanowire field-effect transistor is provided in the ESI,† obtained in all-organic liquid-electrolyte-based inverter circuits
Published on 02 November 2020. Downloaded on 9/15/2023 5:20:46 AM.

Fig. S12. (B11)27 and all-inorganic nanowire inverter circuits (B10).51

Our inverter is a simple integrated circuit consisting of a The switching voltage (Vout = VDD/2) occurs at Vin o 0 rather
pair of nanowire transistors with opposite majority carrier- than the ideal Vin = VDD/2, but this is not unusual for nanowire-
types, as shown in Fig. 3C, designed to take some input signal based inverters.51 The non-ideality arises due to an asymmetry
Vin and invert the waveform, i.e., a maxima/minima in Vin will in channel resistance between the n- and p-type nanowires,51
give a minima/maxima in Vout. The action at a device level is an engineering problem resolvable by tuning the nanowire
illustrated in Fig. 3D; the mechanism is discussed in detail in doping levels or nanowire length ratio to ‘trim’ the circuit into
the literature.24,46,50 A potential difference across the Nafion balance. Data from the individual InAs and GaAs nanowires in
strip drives electric double-layer formation at the electrode- this device as well as DC characterizations of two additional
Nafion and Nafion-nanowire interfaces. Note that although we inverter circuits is provided in ESI,† Fig. S13.
use a metal electrode here, the potential difference could have
an alternative origin, e.g., a neural action potential, as shown AC characterization of inverter performance
for OECT-based neural sensors.11,26 The electric double-layer Fig. 4 shows the AC response of our inverter circuit at a range of
enacts a strong gating action on the nanowire, such that a frequencies obtained with VDD = +1 V. We begin with square-
positive (negative) voltage applied to the input electrode drives wave signals since those are the common test for inverters in
proton accumulation (depletion) at the Nafion-nanowire inter- the literature. We see respectable fidelity at low frequency f,
face. This in turn drives electron accumulation (depletion) with some slight rounding evident at 100 Hz due to capacitive
and enhanced (reduced) conductivity in the InAs nanowire,46 effects. This exacerbates with increasing frequency, as expected,
and hole depletion (accumulation) and reduced (enhanced) since the next edge occurs earlier in the exponential decay that
conductivity in the GaAs nanowire.47 From a circuit perspective, arises from capacitive contributions as f increases. Nonetheless,

Fig. 4 AC signal performance of Nafion-nanowire proton-to-electron transducer (A–D) plots of Vout (top) and Vin (bottom) vs. time at VDD = +1 V for
square-wave input signals at (A) 10 Hz, (B) 100 Hz, (C) 500 Hz and (D) 1 kHz. (E and F) Plots of Vout (top) and Vin (bottom) vs. time at VDD = +1 V for sine-
wave input signals at (E) 1 kHz and (F) 2 kHz.

This journal is © The Royal Society of Chemistry 2021 Mater. Horiz., 2021, 8, 224--233 | 229
 View Article Online

Materials Horizons Communication

we can still distinguish a clear response to an input signal even at Materials and methods
f = 1 kHz, albeit with declining fidelity. Once we get above 1 kHz, Device fabrication
parasitic capacitance contributions from the external circuit begin
to contribute, which could be a significant part of the declining Device substrates were n+–Si wafer (SVMI) coated on the front-
fidelity in Fig. 4A–D. At this point, one needs to remember that a side with 140 nm of thermal SiO2 and on the backside with
square wave is a Fourier series of odd multiples of the funda- Ti/Au to enable the n+–Si to be used as a global back-gate for
mental, e.g., for a 1 kHz spectra, it contains significant compo- devices if desired. The substrates were prepatterned with metal
nents at 3 kHz, 5 kHz, 7 kHz, etc. As such, we reverted to sine wave interconnects by photolithography and alignment markers by
signals in our examples at 1 kHz and 2 kHz in Fig. 4E and F to electron-beam lithography, as is standard for our nanowire
better probe the ability of the Nafion to respond at a specific devices.52 Wurtzite-phase n-type InAs nanowires were grown
frequency. At both frequencies we obtain clean sine wave output, by metal organic vapor phase epitaxy (MOVPE) using Au aero-
albeit with phase shifts of approximately 721 and 941 at 1 kHz and sols with a nominal diameter of 30 nm as seed particles.53 An
2 kHz relative to low frequency operation, respectively, indicating Aixtron 3  2 in2 close-coupled shower-head system was
a growing capacitive contribution to circuit impedance as the operated at a total carrier gas flow of 8 slm and a total reactor
Published on 02 November 2020. Downloaded on 9/15/2023 5:20:46 AM.

frequency is increased. However, in terms of the ability for Nafion pressure of 100 mbar. Growth commenced with a 10 min
to support charge alternation in the electric double-layers anneal in AsH3/H2 ambient at 550 1C with an AsH3 molar
formed at its interfaces, the data in Fig. 4E and F indicates a fraction of wV = 2.5  10 3 before setting the temperature to
response well into the kHz regime, a frequency range of interest 470 1C for nanowire growth. A short InAs stem was grown at
for high-fidelity neural sensing applications,11 that has only very a (CH3)3In molar fraction of wIII = 1.8  10 6 for 180 s with
recently been accessed using PEDOT:PSS OECTs.31 Our work wV = 1.2  10 4 before the wurtzite InAs nanowire was grown
positions Nafion firmly as an alternative material for kHz-range with wV = 2.3  10 5 for 60 min. Growth was terminated by
bioelectronics applications. cutting the (CH3)3In supply and cooling under AsH3/H2 ambient
to 300 1C. The GaAs nanowires were self-catalyzed and grown
by molecular beam epitaxy on (111)Si.54 The undoped core
was grown at 630 1C using As4 and a V/III flux ratio of 60 for
Conclusions 30–45 min. The Be-doped shell was grown at 465 1C using As2
In this work we have demonstrated the nanoscale patterning and a V/III ratio of 150 for 30 min giving nanowires with shell
of Nafion by utilizing it as a negative-tone electron-beam litho- acceptor density NA = 1.5  1019 cm 3, diameter 120  20 nm
graphy resist. This enables direct-write of structures in a scalable and length 5–7 mm. These nanowires are the highest doping
manner, i.e., one only needs to write regions where Nafion should density from earlier work on contacts to p-GaAs nanowires.47,55
remain. After electron-beam exposure, the Nafion is remarkably The nanowires are pure zinc blende crystal phase but may have
resilient to subsequent fabrication processes including three short wurtzite segments at the ends; these wurtzite segments
further rounds of electron-beam lithography, vacuum thermal will be buried under the source/drain contacts. The InAs and
evaporation of metal, and lift-off in hot solvent. The Nafion also GaAs nanowires were dry-transferred to the device substrate
retains excellent protonic properties, and in our devices, the using a custom-built micromanipulator system,43 which
Nafion can respond well to sinusoidal AC signals at frequencies enables a small number of well-separated nanowires to be
as high as 2 kHz. One challenge with Nafion, as we saw in the positioned with sufficient spatial control and then imaged,
device fabrication, is managing its superacid catalyst properties. enabling us to keep track of p-GaAs versus n-InAs in subsequent
For our devices, this required careful ordering of the fabrication processing. We used a 5% solution/suspension of Nafion-117 in
steps. Longer term, in completed devices, it might also present lower aliphatic alcohols and water (a proprietary product by
some issues that need to be managed. From a stability perspec- Sigma-Aldrich, 70160, Lot #BCBW9315). Nafion 117 has an
tive, our devices remain functional after several months of storage equivalent weight of 1100 g mol 1, which corresponds to
under appropriate conditions, e.g., dry nitrogen, which prevents 1100 grams of dry Nafion per mole of sulfonic acid groups
any attack of the Nafion on the nanowires and loss in contact when the material is in the acidic form.38 The solution was used
performance, which is a widely known issue in nanowire devices. either as supplied or diluted up to 1 : 4 in ethanol to give
Finally, we note that the work here opens the path to devices thinner films. Undiluted Nafion solution spin-coated at
beyond the Nafion-nanowire devices we have demonstrated. 3000 rpm for 30 s with 1000 rpm s 1 ramp onto the substrates
One could envision nanoscale artificial synapse devices similar gives a 230 nm film, which is not baked prior to further use.
to those reported at the macroscale by Josberger et al.18 and Nafion film thickness was determined by ellipsometry using a
van de Burgt et al.,19 as well as potentially more complex bioelec- J. A. Woollams M2000 spectroscopic ellipsometer and model-
tronic devices incorporating protonic elements, e.g., protonic ling the Nafion as a Cauchy material following Paul et al.56
transistors.12 There are also possibilities to combine nanoscale- Electron-beam patterning of the Nafion is performed using a
patterned Nafion with ion-porous materials, e.g., PEDOT–PSS, Raith150-Two electron-beam lithography (EBL) system with
assuming methods are found to pattern these at similar scales, beam energy 5–10 keV, 15 mm aperture and dose 3–20 mC cm 2
to overcome challenges in the development of nanoscale bioelec- (typically o10 mC cm 2). The device is then developed in a 1 : 1
tronics. Such advances will be a focus of ongoing work. mixture of 2-propanol : acetone for 60 s at room temperature and

230 | Mater. Horiz., 2021, 8, 224--233 This journal is © The Royal Society of Chemistry 2021
 View Article Online

Communication Materials Horizons

dried. Three EBL-patterned metal depositions then follow to AC inverter characterization

define: (a) ohmic contacts to the InAs nanowires, (b) ohmic Inverter drive bias VDD was applied using a Yokogawa GS200
contacts to the GaAs nanowires, and (c) interconnects to the source-measure unit. An AC input voltage Vin was applied using
Nafion gate structure. The first metallization was vacuum thermal a Stanford Research Systems DS345 signal generator, and the
evaporation of 6 nm Ni and 134 nm Au performed immediately output signal Vout was measured using a Femto DLPVA voltage
after a 60 s immersion in (NH4)2Sx solution at 40 1C to remove the preamplifier and read out using a National Instruments USB-
InAs native oxide. The second metallization was vacuum thermal 6216 Data Acquisition Device with sampling rate 250 kHz. The
evaporation of 160 nm of AuBe alloy (1% Be in Au–ACI Alloys) input signal is also connected to the USB-6216 via a T-piece to
performed immediately after a 30 s immersion in 10% HCl enable simultaneous direct recording of both the applied input
solution to remove the GaAs native oxide. The third metallization and resulting output.
was 2 nm Ti and 60 nm Pd (if Ni/Au is used instead, it is deposited
as part of the first metal deposition step) – the former was tested
to promote absorption of hydrogen from the atmosphere to Author contributions
improve Nafion gating performance, however for this study, we
A. P. M. and P. M. conceived and oversaw the research. J. G. G.
Published on 02 November 2020. Downloaded on 9/15/2023 5:20:46 AM.

found no significant performance gain from Ti/Pd over Ni/Au.

contributed to all aspects of the experimental work, R. L.
developed the initial patterning process for Nafion, J. S.,
Materials characterization K. N., M. L. and F. M. contributed to various aspects of the
XPS was performed with an ESCALAB250Xi X-ray photoelectron fabrication and measurement of the nanowire-Nafion inverters.
spectrometer (Thermo Scientific) with a mono-chromated P. K., J. N. and S. L. contributed to growth of the nanowire
Al Ka-source (energy 1486.68 eV). A beam with 500 mm spot materials, A. B. M. contributed to process development of the
size, 901 photoelectron take-off angle and 120 W power Nafion film, materials/electrical characterization and XPS ana-
(13.8 kV  8.7 mA) was used. Survey and high-resolution scans lysis. All authors provided input during the manuscript pre-
were obtained at 100 eV and 20 eV pass energies respectively. paration and revisions.
Data analysis of the XPS spectra was performed using Shirley
background curves for the high-resolution scans. Peaks were Conflicts of interest
modelled using pseudo-Voigtians with 30% Lorentzian character.
There are no conflicts to declare.
Atmosphere control chamber
All electrical measurements were conducted in an atmospheric
control chamber. The chamber was built from vacuum fittings. Acknowledgements
Dry N2 gas was run through a bubbler filled with deionized
This work was funded by the Australian Research Council (ARC)
water and fed into the chamber inlet to increase the relative
under DP170104024 and DP170102552, the Welsh European
humidity (RH). An outlet on the other side of the system allows
Funding Office (European Regional Development Fund) through
gas to escape to building exhaust. Humidity was monitored
the Sêr Cymru II Program, the Danish National Research Founda-
using a Sensirion SHT2x humidity sensor. The RH saturates at
tion, the Danish Innovation Fund, NanoLund at Lund University,
90–96% after 30–60 min of N2 flow. Unless otherwise specified,
the Swedish Research Council, the Swedish Energy Agency (Grant
all electrical measurements were conducted under these
No. 38331-1) and the Knut and Alice Wallenberg Foundation
conditions.
(KAW). P. M. is a Sêr Cymru Research Chair and an Honorary
Professor at the University of Queensland and A. B. M. is a Sêr
DC single nanowire characterization Cymru II fellow and the results incorporated in this work have
Source–drain bias VSD was applied using a Keithley K2401 received funding from the European Union’s Horizon 2020
source-measure unit. The resulting drain current ID was mea- research and innovation program under the Marie Skłodowska-
sured with a Keithley 6517A electrometer. The gate voltage VG Curie grant agreement no. 663830. A. P. M. was a Japan Society for
was applied using a Keithley K2400 source-measure unit the Promotion of Science (JSPS) Long-term Invitational Fellow
enabling gate leakage current monitoring. during the drafting of this manuscript. The work was performed
in part using the NSW and Queensland nodes of the Australian
DC inverter characterization National Fabrication Facility (ANFF) and the Electron Microscope
Unit (EMU) within the Mark Wainwright Analytical Centre
Inverter drive bias VDD was applied using a Keithley K2401
(MWAC) at UNSW Sydney.
source-measure unit. A DC input voltage Vin was applied using a
Keithley K2400 enabling gate leakage current monitoring,
and the output voltage Vout was measured using a Femto DLPVA Notes and references
voltage preamplifier and read out with a Keithley 2000 multi-
meter. The inverter drive was grounded via a Keithley 6517A 1 J. Rivnay, H. Wang, L. Fenno, K. Deisseroth and G. G.
electrometer to enable monitoring of the inverter drive current. Malliaras, Sci. Adv., 2017, 3, e1601649.

This journal is © The Royal Society of Chemistry 2021 Mater. Horiz., 2021, 8, 224--233 | 231
 View Article Online

Materials Horizons Communication

2 T. Someya, Z. Bao and G. G. Malliaras, Nature, 2016, 540, 26 D. Khodagholy, J. Rivnay, M. Sessolo, M. Gurfinkel,
379–385. P. Leleux, L. H. Jimison, E. Stavrinidou, T. Herve,
3 A. Noy, Adv. Mater., 2011, 23, 807–820. S. Sanaur, R. M. Owens and G. G. Malliaras, Nat. Commun.,
4 J. Rogers, G. Malliaras and T. Someya, Sci. Adv., 2018, 2013, 4, 2133.
4, eaav1889. 27 H. Sun, M. Vagin, S. Wang, X. Crispin, R. Forchheimer,
5 G. D. Spyropoulos, J. N. Gelinas and D. Khodagholy, Sci. M. Berggren and S. Fabiano, Adv. Mater., 2018, 30, 1704916.
Adv., 2019, 5, eaau7378. 28 A. Giovannitti, C. B. Nielsen, D.-T. Sbircea, S. Inal,
6 R. M. Owens and G. G. Malliaras, MRS Bull., 2010, 35, 449–456. M. Donahue, M. R. Niazi, D. A. Hanifi, A. Amassian,
7 J. Rivnay, R. M. Owens and G. G. Malliaras, Chem. Mater., G. G. Malliaras, J. Rivnay and I. McCulloch, Nat. Commun.,
2014, 26, 679–685. 2016, 7, 1–10.
8 M. Berggren and G. G. Malliaras, Science, 2019, 364, 29 Y. Sone, P. Ekdunge and D. Simonsson, J. Electrochem. Soc.,
233–234. 1996, 143, 1254.
9 J. Rivnay, S. Inal, A. Salleo, R. M. Owens, M. Berggren and 30 J. Selberg, M. Jia and M. Rolandi, PLoS One, 2019,
G. G. Malliaras, Nat. Rev. Mater., 2018, 3, 17086. 14, e0202713.
Published on 02 November 2020. Downloaded on 9/15/2023 5:20:46 AM.

10 F. Jonas, W. Krafft and B. Muys, Macromol. Symp., 1995, 100, 31 C. Cea, G. D. Spyropoulos, P. Jastrzebska-Perfect,
169–173. J. J. Ferrero, J. N. Gelinas and D. Khodagholy, Nat. Mater.,
11 D. Khodagholy, T. Doublet, P. Quilichini, M. Gurfinkel, 2020, 19, 679–686.
P. Leleux, A. Ghestem, E. Ismailova, T. Hervé, S. Sanaur, 32 A. Omosebi and R. S. Besser, J. Electrochem. Soc., 2011, 158,
C. Bernard and G. G. Malliaras, Nat. Commun., 2013, 4, 1575. D603–D610.
12 C. Zhong, Y. Deng, A. F. Roudsari, A. Kapetanovic, M. P. 33 A. Omosebi and R. S. Besser, J. Power Sources, 2013, 228,
Anantram and M. Rolandi, Nat. Commun., 2011, 2, 476. 151–158.
13 Y. Deng, E. Josberger, J. Jin, A. F. Roudsari, B. A. Helms, 34 A. M. Steinbach, S. Jenisch, P. Bakhtiarpour, M. Amirkhani
C. Zhong, M. P. Anantram and M. Rolandi, Sci. Rep., 2013, and S. Strehle, MRS Adv., 2016, 1, 45–50.
3, 2481. 35 K. Li, J. Li, C. Reardon, C. S. Schuster, Y. Wang, G. J. Triggs,
14 K. Tybrandt, R. Forchheimer and M. Berggren, Nat. Commun., N. Damnik, J. Müenchenberger, X. Wang, E. R. Martins and
2012, 3, 871. T. F. Krauss, Sci. Rep., 2016, 6, 32945.
15 M. Berggren, X. Crispin, S. Fabiano, M. P. Jonsson, D. T. 36 S. C. DeCaluwe, A. M. Baker, P. Bhargava, J. E. Fischer and
Simon, E. Stavrinidou, K. Tybrandt and I. Zozoulenko, Adv. J. A. Dura, Nano Energy, 2018, 46, 91–100.
Mater., 2019, 31, 1805813. 37 N. Pala and M. Karabiyik, in Encyclopedia of Nanotechnology,
16 D. Mawad, C. Mansfield, A. Lauto, F. Perbellini, G. W. ed. B. Bhushan, Springer, Netherlands, Dordrecht, 2016,
Nelson, J. Tonkin, S. O. Bello, D. J. Carrad, A. P. Micolich, pp. 1033–1057.
M. M. Mahat, J. Furman, D. Payne, A. R. Lyon, J. J. Gooding, 38 K. A. Mauritz and R. B. Moore, Chem. Rev., 2004, 104,
S. E. Harding, C. M. Terracciano and M. M. Stevens, Sci. 4535–4586.
Adv., 2016, 2, e1601007. 39 R. B. Moore and C. R. Martin, Macromolecules, 1988, 21,
17 E. Stavrinidou, R. Gabrielsson, E. Gomez, X. Crispin, 1334–1339.
O. Nilsson, D. T. Simon and M. Berggren, Sci. Adv., 2015, 40 H. J. Levinson, Principles of Lithography, SPIE Press, Bellingham,
1, e1501136. WA, 2nd edn, 2005.
18 E. E. Josberger, Y. Deng, W. Sun, R. Kautz and M. Rolandi, 41 Q. Sun, K. Hu, K. Leng, X. Yi, B. Aguila, Y. Sun, A. Zheng,
Adv. Mater., 2014, 26, 4986–4990. X. Meng, S. Ma and F.-S. Xiao, J. Mater. Chem. A, 2018, 6,
19 Y. van de Burgt, E. Lubberman, E. J. Fuller, S. T. Keene, 18712–18719.
G. C. Faria, S. Agarwal, M. J. Marinella, A. Alec Talin and 42 M. DeJarld, J. C. Shin, W. Chern, D. Chanda, K. Balasundaram,
A. Salleo, Nat. Mater., 2017, 16, 414–418. J. A. Rogers and X. Li, Nano Lett., 2011, 11, 5259–5263.
20 X. Strakosas, J. Selberg, X. Zhang, N. Christie, P. Hsu, 43 J. Seidl, J. G. Gluschke, X. Yuan, S. Naureen, N. Shahid,
A. Almutairi and M. Rolandi, Adv. Sci., 2019, 6, 1800935. H. H. Tan, C. Jagadish, A. P. Micolich and P. Caroff, Nano
21 X. Chen, H. Zhang, R. H. Tunuguntla and A. Noy, Nano Lett., Lett., 2019, 19, 4666–4677.
2019, 19, 629–634. 44 J. K. Lim, B. Y. Lee, M. L. Pedano, A. J. Senesi, J.-W. Jang,
22 Z. Hemmatian, R. H. Tunuguntla, A. Noy and M. Rolandi, W. Shim, S. Hong and C. A. Mirkin, Small, 2010, 6,
PLoS One, 2019, 14, e0212197. 1736–1740.
23 R. H. Tunuguntla, M. A. Bangar, K. Kim, P. Stroeve, 45 J. G. Gluschke, J. Seidl, A. M. Burke, R. W. Lyttleton,
C. Grigoropoulos, C. M. Ajo-Franklin and A. Noy, Adv. D. J. Carrad, A. R. Ullah, S. Fahlvik, S. Lehmann, H. Linke
Mater., 2015, 27, 831–836. and A. P. Micolich, Nanotechnology, 2018, 30, 064001.
24 D. J. Carrad, A. B. Mostert, A. R. Ullah, A. M. Burke, H. J. Joyce, 46 D. J. Carrad, A. M. Burke, R. W. Lyttleton, H. J. Joyce,
H. H. Tan, C. Jagadish, P. Krogstrup, J. Nygård, P. Meredith H. H. Tan, C. Jagadish, K. Storm, H. Linke, L. Samuelson
and A. P. Micolich, Nano Lett., 2017, 17, 827–833. and A. P. Micolich, Nano Lett., 2014, 14, 94–100.
25 J. Rivnay, P. Leleux, M. Sessolo, D. Khodagholy, T. Hervé, 47 A. R. Ullah, D. J. Carrad, P. Krogstrup, J. Nygård and
M. Fiocchi and G. G. Malliaras, Adv. Mater., 2013, 25, 7010–7014. A. P. Micolich, Phys. Rev. Mater., 2018, 2, 025601.

232 | Mater. Horiz., 2021, 8, 224--233 This journal is © The Royal Society of Chemistry 2021
 View Article Online

Communication Materials Horizons

48 D. Liang and X. P. A. Gao, Nano Lett., 2012, 12, 3263–3267. 53 S. Lehmann, J. Wallentin, D. Jacobsson, K. Deppert and
49 M. Sheliakina, A. B. Mostert and P. Meredith, Mater. Horiz., K. A. Dick, Nano Lett., 2013, 13, 4099–4105.
2018, 5, 256–263. 54 A. Casadei, P. Krogstrup, M. Heiss, J. A. Röhr, C. Colombo,
50 S. H. Kim, K. Hong, W. Xie, K. H. Lee, S. Zhang, T. P. Lodge T. Ruelle, S. Upadhyay, C. B. Sørensen, J. Nygård and
and C. D. Frisbie, Adv. Mater., 2013, 25, 1822–1846. A. Fontcuberta i Morral, Appl. Phys. Lett., 2013, 102, 013117.
51 A. W. Dey, J. Svensson, B. M. Borg, M. Ek and L.-E. 55 A. R. Ullah, J. G. Gluschke, P. Krogstrup, C. B. Sørensen,
Wernersson, Nano Lett., 2012, 12, 5593–5597. J. Nygård and A. P. Micolich, Nanotechnology, 2017, 28, 134005.
52 K. Storm, G. Nylund, L. Samuelson and A. P. Micolich, Nano 56 D. Paul, A. Fraser, J. Pearce and K. Karan, ECS Trans., 2011,
Lett., 2012, 12, 1–6. 41, 1393–1406.
Published on 02 November 2020. Downloaded on 9/15/2023 5:20:46 AM.

This journal is © The Royal Society of Chemistry 2021 Mater. Horiz., 2021, 8, 224--233 | 233